The efficacy of Heimberg's (1991) Cognitive-Behavioral Group Therapy (CBGT) [Unpublished manuscript] for social phobia has been demonstrated in several studies in recent years. However, little is known about the mechanisms underlying the treatment's success. In order to determine whether the cognitive restructuring component of CBGT is essential, this study compared CBGT to an exposure-based treatment without formal cognitive restructuring. A wait-list control was also included. In general, Ss in the active treatment conditions improved and control Ss did not improve on a variety of self-report, clinician, and behavioral measures. Limited evidence indicated that Ss in the non-cognitive treatment may have made somewhat greater gains on some measures. Although CBGT Ss reported more improvement than exposure-alone Ss in subjective anxiety during an individualized behavioral test at posttreatment, this difference disappeared at 6-month follow-up. Surprisingly, CBGT was less effective than in previous controlled trials, and possible reasons for this are discussed. Implications of the results for cognitive theory and cognitive-behavioral therapy for social phobia are addressed.